Any idea how DN search works?
Maybe it's better to ask how can DN's search be improved? When I search for an article I read last week on Figma, I'd prefer to see the most recent articles first–not from over a year ago.
There have been many times I've wanted to re-find something I saw on Designer News only to have the search turn up old/unrelated results.
Same happens to me. If you happen to think of something you want to go back and look at, it's virtually impossible to find it again. It's pretty horrible
I think It doesn't matter much how question should be asked in this case. Simple its so bad
It's great! You go to Google.com, then type
"[your query] designernews"
"[your query]" site:designernews.co
Would it still include news.layervalult.com articles though?
"[your query]" site:designernews.co OR site:news.layervault.com
Yes I think. They moved domain name on google as far as I know
Agree. its pretty bad.
I completely agree. Some of the things I dislike are the timestamp, the lack of space between the results and the lack of sorting by.
To be more specific, a result shows something like "over 3 years ago" or "almost 1 year ago", "about 4 years ago" which is useless, just write "3 years ago", no need to add more text that delays the reading.
Also, the result blocks don't have enough padding between each other so you don't know if "STORY" is the title of the block above or below, and the ubiquitous thumbnail doesn't help.
The search page says that it's powered by Swiftype, maybe DN should change service. There's no arguing that an improvement is really needed!
Full disclosure: I work at Swiftype, the company that powers DN's search.
I am certainly biased, but I'd like to think that the quality of the search experience on DN has little to do with the underlying technology and a lot to do with the implementation. I think some relatively simple changes would make a huge difference:
- Adding some basic facets, like badges and date range so that users can narrow their searches quickly.
- Consider tuning the relevance algorithm to weight Stories over Comments and Users, or possibly filtering to Stories by default.
- Consider sorting Stories by overall comment activity or other popularity metrics, like clicks.
- Adding more metadata to the results themselves, like the user who posted, the number of comments, and an excerpt if available.
If anyone from DN reads this, I'd be happy to help if you've got any questions.
Honestly, just return results for the keywords entered by latest to oldest. The search returns things that are old and seems to not index more recent stuff as well. Or just not return things that relevant at all.
Date bucketing may help, but it's also an indication of a return failure (search -> result -> expectation not met -> ok i'll try filter). There is a 0.1% chance someone will remember who posted an article.
I don't know about search, never use it, but the notification system has been broken for years. I haven't been able to see a recent list of comments and replies. I get s notification dot but if I want to go back and reread the comments I can't. Well, not without using the broken search feature apparently.
This is why it's not good: There is no sort. It's sorted by some mysterious algorithm and a user has no control over it. Therefore, there is little to no context to what you are looking at.
The filters they brought it are confusing as well. "Everything else" <- What does that mean?
Edit: I just discovered the sorting mechanism. You have to filter first. Head bang
Maybe it would be worth it to move it to Algolia?
Until it is improved...Choose search type as "story" and then click "newest"